Fine-Needle Aspiration Biopsy of the Lymph Node: A Novel Tool for the Monitoring of Immune Responses after Skin Antigen Delivery

This article is featured in In This Issue, p.1

The skin is widely used as a convenient access route to the immune system. Intradermal administration of Ags is used both to induce responses (as in vaccination) and to monitor immune memory (as in delayed type hypersensitivity tests) as well as in attempts to induce Ag specific tolerance (1, 2). The immense potential of the dermal route has been underlined in recent years by the discovery that the epidermis and upper dermis not only have a very high density and wide variety of APCs (3, 4) but also a large previously underappreciated reservoir of resident T cells, twice as large as the blood pool (5). This reservoir is polyclonal and in contrast to the blood or lymph node (LN) contains almost exclusively T cells of memory phenotype (6). Recent studies are consistent with the view that these cells represent a very comprehensive and long-lived memory “library” not only of Ag exposure that has previously occurred in that area of skin but also of responses that have occurred elsewhere in the skin and probably in other epithelial barrier tissues as well (7). This is achieved by the generation of memory T cells with skin homing receptors during immune responses occurring at other locations, which recirculate to the skin all over the body and reside there for many years. In this study, these T cells represent a first line of adaptive immunity against future antigenic challenge occurring anywhere across the body surface (8). T cells specialized for entry into cutaneous sites are characterized by expression of high levels of cutaneous leukocyte Ag (CLA) (9). They are predominantly CD45RA⁻CCR7⁻ effector memory cells, but there is a proportion of CLA⁺CCR7⁺ cells, which are known as “central memory skin homing cells” that can enter the skin but most likely have access to the LNs as well (5). A subpopulation of CD45RA⁻CCR7⁻ cells, also called peripheral tissue immune surveillance T cells, express CCR8, which regulates their localization in the skin during steady state, whereas other skin homing markers, such as CCR4 and CCR10, are largely responsible for the localization of activated T cells during immune responses (10).

This system is well illustrated by the Mantoux reaction in which previous exposure to Mycobacaterium tuberculosis, for example in the lung, generates tuberculin purified protein derivative (PPD)–reactive memory cells that become resident in uninvolved areas of the skin and for years afterward can respond within 48 h to intradermal Ag challenge (11–13).

However, the skin (and other peripheral tissues) is not the only site of storage of long-term T cell memory. The blood represents a store of both central and effector memory T cells, available immediately to all parts of the body. CD45RA⁻CCR7⁻ cells, or effector memory cells, are common in blood but rare in the LN because of the lack of CCR7. They are mainly bound for inflamed peripheral tissues and rapidly produce inflammatory cytokines such as IFN-γ, IL-4, or IL-5 (14).

By contrast, lymphoid tissue stores mainly central memory (CD45RA⁻CCR7⁺), which together with LN-resident naive T cells (CD45RA⁺CCR7⁺) (15), allow the LN not only to rapidly expand the effector memory T cell pool during a recall response (16) but also to prime the naive repertoire during a primary immune response (17).

To understand the interplay between these sites (the skin, the draining LN, and the blood), it is important to be able to sample them before, during, and after immune responses. In humans, blood is readily accessible, and techniques such as skin biopsy and skin blister formation have been developed to sample the skin repertoire (18). However, although LN sampling is routine in animals, to date, techniques have not been developed to do so in humans. This is an important technique to develop not only to provide a means of monitoring central memory in this compartment but also because draining LNs are apparently the only exit route for T cells from the skin and hence represent a potentially sensitive means of monitoring how the peripheral tissues communicate with the rest of the body during an immune response.

For this purpose, we have used minimally invasive ultrasound-guided fine-needle aspiration (FNA) of the LN, a technique commonly used in clinical practice for cancer diagnosis (19). Axillary LN aspiration was used to sample the draining LN following PPD administered intradermally in the forearm.

The aims of this study were to establish the following: 1) whether ultrasound-guided LN FNA in humans be used to reliably obtain normal LN tissue; 2. is LN FNA and repeat sampling an acceptable and well tolerated procedure in healthy volunteers; 3) can ultrasound-guided FNA of draining LNs be used to detect Ag-specific responses following tissue administration of Ag in humans; and 4) how do LN responses differ from those measured in the peripheral blood?

The study was conducted in accordance with International Conference on Harmonization/World Health Organization Good Clinical Practice standards. All subjects provided written informed consent prior to enrolment in the study, according to the Declaration of Helsinki (revision 2013).

This study, fully approved by South East Wales Research Ethics Committee, was a single center, before-after study of the optimization of the detection of the immune response to intradermally delivered tuberculin PPD in healthy volunteers (4 males and 21 females), ages 18–50 y.

Participants had pretreatment blood sample and axillary LN FNA biopsy, followed by injection of a single dose of 2 tuberculin units (TU) PPD (Serum Statens Institute, Copenhagen, Denmark) by a hollow microneedle (Micronjet 600 Needle; Nanopass Technologies, Israel), which was previously used for vaccination in various clinical studies (20). The injection was administered intradermally in the forearm, contralateral to the initial LN FNA. Participants with a positive reaction (induration at the site of tuberculin PPD injection of >5 mm in diameter, 48 h after injection) had a posttreatment blood sample and axillary LN FNA, ipsilateral to the injection site, 2–5 d after injection.

Under aseptic conditions, the skin and s.c. tissues down to the identified LN were infiltrated by 1–2 ml 1% (w/v) xylocaine with 1:200,000 adrenaline. Under real-time ultrasound visualization, using a 21-gauge needle and a 5-ml syringe, the LN cortex was sampled using the FNA technique of small to and fro needle movements while applying 1 ml suction with the syringe. By this technique, a wide area of the LN cortex was sampled. One to two passages were used per sample. The LN sample was immediately transferred to medium—10% (v/v) human AB serum (PAA, Pasching, Austria) in RPMI 1640 medium (Life Technologies, Paisley, U.K.)—by “washing” the FNA needle several times with the medium (19) and transported to the laboratory at ambient temperature. The number of white cells in the LN sample was counted in a hemocytometer after staining with trypan blue (1:3 dilution in PBS) and 4% (v/v) acetic acid to eliminate RBC contamination. All samples were used fresh on the day of collection.

PBMCs were isolated on Ficoll–Paque Plus (1.077 ± 0.001 g/ml, +20°C; GE Healthcare Biosciences, Sweden) gradient centrifugation from the peripheral blood of healthy donors, taken at the same time as LN samples. All samples were used fresh on the day of collection.

The following mAbs were used: PacificBlue-anti-human-CD3 (clone S4.1; Invitrogen, Paisley, U.K.), eFluor 506 viability dye (eBioscience, Hatfield, U.K.), biotin-anti-human-CCR8 [a gift from the Prof. B. Moser group, Cardiff University, Cardiff, U.K. (21)] with anti-biotin PE (clone 1D4-C5), FITC-anti-human-CLA (clone HECA-452), PE-anti-human-CCR7 (clone G043H7), allophycocyanin-anti-human-CD8 (clone HIT8a), allophycocyanin-anti-human CCR10 (clone 314305), PerCPCy5.5-anti-human-CD69 (clone FN50), PerCPCy5.5-anti-human-CD8 (clone SK1), PeCy7-anti-human-CD45RA (clone HI100), PeCy7-anti-human-CCR4 (clone L291H4), and allophycocyanin-Cy7 anti-human CD4 (clone OKT4) (all from BioLegend, London, U.K.). The cells were analyzed on FACSCanto II flow cytometer (BD Biosciences) and using FlowJo software version 8.8.6 (Leland, Stanford, U.K.).

We used ELISPOT for the detection of IFN-γ–producing CD4 cells (blood and LN derived) in response to tuberculin PPD as described previously (22, 23). PBMCs and LN cells (LNCs) were cultured in polypropylene round-bottomed tubes in 10% (v/v) human AB serum in RPMI 1640 supplemented with antibiotics (Antibiotic-Antimycotic; Sigma-Aldrich, Poole, U.K.) and tuberculin PPD (0.05 and 0.5 μg/ml) or control Ags (tetanus toxoid at 0.5 U/ml [Novartis, Holzkirchen, Germany] and viral peptide mix containing flu matrix protein (MP)_58–66 [GILGFVFTL], CMV pp65_495–503 [NLVPMVATV], and EBV BMLF1_280–288 [GLCTLVAML] at 20 nM each [GL Biochem, Shanghai, China]) at 37°C and 5% (v/v) CO₂.

Fifteen to 18 h later, the cells were washed and resuspended using prewarmed AIM-V+AlbuMAX BSA serum-free medium (Life Technologies, Paisley, U.K.) and brought to concentration of 10⁵/100μl for LNCs and 10⁶/100 μl for PBMCs, after which 100μl/well was dispensed in two to three replicate wells (depending on cell availability) of ELISPOT filter plates (Millipore, Billerica, MA) preblocked with 1% (w/v) BSA (Promega, Madison, WI) in PBS (Sigma-Aldrich) and precoated with anti–IFN-γ capture Ab (U-CyTech, Utrecht, Netherlands). Control wells contained PBMCs only or PBMCs cultured with PHA (Sigma-Aldrich) at final concentration of 1 μg/ml in a 100-μl volume. After capture at 37°C, 5% CO₂ for 18 h, plates were washed manually: three times with PBS, followed by 0.05% (v/v) Tween 20 (Fisher-Scientific, Loughborough, U.K.) in PBS. Spots were then developed in steps involving incubation with biotinylated anti–IFN-γ detection Ab (U-CyTech), ExtrAvidin-Alkaline Phosphatase (Sigma-Aldrich) and 5-bromo-4-chloro-3-indolyl phosphate/NBT (Sigma-Aldrich) as a substrate for the alkaline phosphatase activity.

Plates were analyzed on a BioReader (ByoSys, Karben, Germany) as follows. First, the spots were counted as total number of spots per condition, summing all available wells and using customized settings in the BioReader (only spots with diameter ≥ 40 μm were taken into account). Second, the number of background spots (spots from wells containing cells not exposed to Ag) was subtracted from each condition. Third, the results are expressed as number of IFN-γ–producing cells/2 × 10⁵ cells.

Paired sample t test or Wilcoxon signed rank test was performed to test the differences between paired measurements. For comparison of three or more samples, one-way ANOVA was used. Analysis was performed by using Prism GraphPad 4.0a for Macintosh. A p value < 0.05 was considered significant. Results were expressed as mean ± SD.

We obtained on average of 3.3 ± 2.1 × 10⁶ (range 1-10 × 10⁶) white cells per LN aspirate, calculated from 39 analyzed samples, collected from 25 healthy volunteers (21 females and 4 males) with the mean age of 28.9 ± 7.1 y.

Although the predominance of CD4⁺ cells was obvious in both blood-derived (CD4: 61.9 ± 8.1%, CD8:26.5 ± 5.3%; ratio 2.33) and LN-derived CD3⁺ cells (CD4: 79.6 ± 9.9%, CD8: 15.3 ± 6.7%; ratio 5.20), this was significantly more marked in the LN aspirate (n = 12, p = 0.0001) (Fig. 1A).

We analyzed four matching LN and blood samples for the expression of CD45RA and CCR7 to establish the distribution of naive and memory T cells. The CD45RA⁺CCR7⁺ cells were characterized as naive, CD45RA⁺CCR7⁻ as T_EMRA cells (i.e., effector memory RA⁺ T cells) CD45RA⁻CCR7⁺ as central memory cells and CD45RA⁻CCR7⁻ cells as effector memory cells (24). There were significantly fewer CD8⁺ T_EMRA cells in the LN samples, even when the CD8⁺ population only, was taken into account (percentage of CD3⁺: 0.92 ± 0.88 in LNCs versus 10.66 ± 6.53 in PBMCs, p = 0.04; percentage of CD8⁺: 3.37 ± 1.93 in LNCs versus 22.53 ± 7.65 in PBMCs, p = 0.01). We also observed a trend toward more marked predominance of central to effector memory in the LN, again more consistent in the CD8⁺ population (Fig. 1B).

Analysis of CD69 expression revealed a significantly higher proportion of CD69⁺ cells in the LN, in both the CD4⁺ and CD8⁺ populations. Overall percentage of CD69⁺CD3⁺ was markedly increased among LNCs in comparison with PBMCs (27.67 ± 7.49 versus 3.49 ± 2.62%, respectively; n = 7, p < 0.0001).

Analyzing the combined expression of CD45RA and CD69, different populations of T cells were identified: 1) CD69⁻ naive cells, 2) CD69⁺ naive cells, 3) CD69⁺ memory, and 4) CD69⁻ memory, allowing for the limitation of such a gating strategy, particularly having in mind that some of the CD45RA⁺ cells in both CD3⁺ and CD8⁺ populations are T_EMRA cells, rather than naive T cells. Percentage of CD69⁻ cells, both naive and memory, was significantly higher in blood, whereas CD69⁺ naive and memory T cells were significantly more represented in the LN. This similar pattern was observed in CD3⁺, CD4⁺, and CD8⁺ populations (Fig. 2).

Analysis of the expression of skin-homing receptors in the blood and the LN showed no significant differences, with the exception of significantly higher expression of CCR4 and CLA in the CD8⁺ blood cells in comparison with the LN-derived cells (p = 0.01 and p = 0.03, respectively) as illustrated in Fig. 3A. Interestingly, the majority of CLA⁺CD3⁺ cells in the blood and the LN were central memory cells (n = 4; p = 0.02 and p = 0.007, respectively) (Fig. 3B)_. A similar pattern was seen in the CD4⁺ and CD8⁺ populations (data not shown).

Next, we proceeded with the detection of the immune response in the LN after in vivo tuberculin PPD injection. Of 22 participants who received tuberculin PPD, 15 were Mantoux positive according to the criteria defined above.

We observed a significant increase in the ultrasonographically measured thickness of the LN cortex 2 d after tuberculin PPD injection (before: 0.25 ± 0.22 cm; 2 d after: 0.35 ± 0.35 cm; p = 0.007, n = 10) but not after 5 d (before: 0.20 ± 0.04 cm; 5 d after: 0.22 ± 0.06 cm; p = 0.72, n = 5) (Fig. 4).

To detect the response in the regional axillary LN, after in vivo intradermal injection of 2 TU tuberculin PPD in the forearm, we used an ELISPOT assay designed to detect IFN-γ–producing cells among LNCs and PBMCs collected before and after the injection. It involved preincubation of cells with an increasing concentration of tuberculin PPD (0.5 and 5 μg/ml). Stimulants, such as1 μg/ml PHA and 0.5 TU/ml tetanus toxoid, were used as controls.

Representative examples of ELISPOT plates demonstrating a response of LNCs and PBMCs to the increasing in vitro concentrations of tuberculin PPD and control stimulants, before and after in vivo tuberculin PPD injection, are shown in Fig. 5.

There was a significant increase in IFN-γ–producing cells among LNCs after two days of tuberculin PPD injection (n = 7, p = 0.008), whereas the response in PBMCs remained unchanged (n = 7, p = 0.22). At five days, no difference compared with baseline was seen with LNCs (n = 5, p = 0.25); the response in PBMCs was marginally decreased at this time point (n = 5, p = 0.03) (Fig. 6A, 6B).

We observed an increase in the response of LNCs to PHA following in vivo tuberculin PPD injection (Fig. 6C). However, no difference in the response to control Ags—tetanus toxoid and mixed viral peptides—was seen (Fig. 6D).

Our results suggest that ultrasound-guided aspiration represents a well-tolerated means of minimally invasive draining LN sampling in humans. Although such sampling requires a skilled radiologist/ultrasonographer, in our hands, repeat sampling as little as 2 d apart is feasible, provides at least 10⁶ white cells, and has a very low failure rate (<5%, no failures and no postprocedural complications seen in 25 aspirates). We are aware of one previous report of axilliary LN sampling for research purposes in which no comment was made on complications and only limited flow cytometry was performed (25).

The flow cytometric profile of the cells obtained from LN sampling was distinct from the profile of paired PBMCs from the same subjects in several ways. In particular, LN cells displayed a higher CD4/CD8 ratio (5.20 versus 2.33, p = 0.0001), fewer effector memory cells, absent CD8⁺ T_EMRA cells (3.37 ± 1.93 in LNCs versus 22.53 ± 7.65 in PBMCs, p = 0.01), and markedly increased levels of CD69 on all naive and memory T cell subsets (LNCs: 27.67 ± 7.49% versus PBMCs: 3.49 ± 2.62%, n = 7, p < 0.0001), which is most likely to represent a retention, rather than activation marker in this tissue. The LN profile seen on aspiration was very consistent with the populations reported in postmortem and surgical samples (26).

A remarkable finding was the low response to PPD seen in LN prior to Mantoux testing. This very low “background” was all the more surprising because PBMCs from the same individuals showed very marked ELISPOT responses at the same time point. The contrast was evident even if we allow for the difference in the number of PBMCs and LNCs plated in the wells (2 × 10⁶ PBMCs/well, 1 × 10⁶ LNCs/well). The low background response in LNs (in the absence of skin challenge) might then be explained by the predominance of central memory cells in LNs—tending to express IL-2 rather than IFN-γ (14). However, 48 h after tuberculin PPD administration intradermally, a clear increase in IFN-γ ELISPOT counts was seen in all individuals (after/before ratio: 14.34 ± 13.08).

After 5 days, the PPD-specific response was no longer different from baseline in the LN. This contrasts with time course of Ag-specific T cell proliferation in the skin after cutaneous Ag challenge described by Akbar and colleagues (27), which begins to appear at 2 d and peaks at 7–10 d. Note, however, that the Mantoux skin reaction itself appears characteristically by 48 h. In our view, these apparently discrepant findings are best explained by the skin effector memory T cell subset being responsible for both the 48-h IFN-γ response in the LN (via rapid migration to the LN) as well as the local Mantoux reaction in skin at 48 h. The strong response, although unchanged from baseline, seen in PBMCs at this time point is then also explicable by the significantly greater numbers of effector memory cells present in blood (Fig. 1B). The later accumulation of cells in the skin described between 2 and 10 d post-Mantoux is likely to represent a combination of recruitment of T cells from the blood following local skin inflammation and proliferation of central memory T cells also present in the skin (27).

Evidence from murine experiments and studies of T cell depletion in human skin T cell malignancies (leukemic CTCL and mycosis fungoides) suggest that central memory T cells from skin regularly recirculate, whereas this is not the case for skin effector memory cells (28, 29). However, the possibility of migration to the draining LN (but not recirculation to other tissues) following Ag stimulation was not excluded, especially for CD4⁺ T cells, and indeed, it is reported that some patients with malignant disease of effector memory T cells (mycosis fungoides) can develop LN involvement (28). Our data confirm that the majority of skin-tropic T cells in the blood during noninflammatory conditions are of the central memory type. We also showed a similar finding in the LN, indicating that the blood-skin-LN path is dominated by central memory cells in the steady (non-Ag simulated) state.

A potential concern in the data presented in this paper is a “nonspecific” element in the increase in ELISPOT responses post-skin challenge. Enlargement of LNs was detectable (Fig. 4) with a 3-fold increase in the PHA response (Fig. 6). However, this did not result in an increased (unstimulated) background response to PPD or in the appearance of responses to Ags other than PPD (e.g., tetanus, “viral mix”), even when a strong peripheral blood recall response to these Ags was present in the same individual (Fig. 5). This emphasizes the importance of including control Ags (and not just general mitogens) in such assays.

Our findings have several important implications. Although ultrasound-guided aspiration can reliably access LN cells, IFN-γ ELISPOTs—at least with short time course stimulation (<24 h)—in the absence of skin challenge cannot be used to probe “resting” immune memory, because no response was seen when there was clearly PPD memory present in the peripheral blood (Figs. 5, 6D). Different protocols may be required to examine the central memory repertoire in LN cells—either prolonged stimulation, or the use of different indicator cytokines such as IL-2. However, the lack of effector memory T cells in LNs appears to make this site very suitable to detect the immediate T cell response to Ag challenge in the skin, presumed mediated by migrating effector T cells. Hence, it appears that the combination of intradermal Ag administration and draining LN sampling can be used to probe the effector memory T cell repertoire in the skin (Fig. 6A). By contrast, changes in the blood level of responding T cells were not detectable even 5 d after intradermal PPD challenge (Fig. 6B).

In conclusion, we have shown that ultrasound-guided FNA can be used to study skin-draining LN cells in humans. In addition to the testing of the delayed hypersensitivity response in the LN, shown in this study, this approach has a potential for monitoring of the response to less potent Ags and alternative aspects of memory such as recall regulatory T cell responses (30). Further studies using different antigenic challenges, cytokine readouts, and time points as well as simultaneous skin sampling will be required to confirm the exact nature and time course of the cellular signaling between skin and draining LNs.

We thank Prof. Bernard Moser and Dr. Michelle McCully for helpful discussion and providing the biotin-anti-human-CCR8 Ab.

Y.L. and E.K. are employees of NanoPass Technologies Ltd., the provider of the MicronJet 600 device.